Research Internship: ROI masking in a simulated Teleoperated Driving Environment

Speaker: Fuqi Guan

Room: 1967

For fully immersive telepresence within a vehicle a 360° Field of View is required. As the human FoV is significantly smaller than 360°, just a slice of the surround view has to be transmitted to the operator. Within the operator’s current Field of View, there are regions in which the operator is more interested in than in others. Especially within a vehicular context these Regions of Interest (ROI) remain almost unchanged during one session. The objective of this thesis is to set up a teleoperation driving environment based on a video game e.g. GTA V [2], [3] or driving simulation e.g. Carla [4] and analyze the influence of ROI adaption on the operator’s performance. ROI adaptions (single color, blur, content based, etc.) as suggested by [1] should be performed on the non-interesting parts of the video frame and evaluated regarding the operator’s performance, QoE and total rate reduction.

[1] Furman, Vadim, Andrew Hughes Chatham, Abhijit Ogale, und Dmitri Dolgov. Image and video compression for remote vehicle assistance. United States US9767369B2, filed 3. Juni 2016, und issued 19. September 2017.

[2] github.com/aitorzip/DeepGTAV

[3] youtu.be/StR9tuaIwYw

[4] carla.org